Cosmological constraints from the DESI DR1 joint power spectrum and bispectrum analysis

Abstract

We derive cosmological parameter constraints from the Dark Energy Spectroscopic Instrument (DESI) Data Release 1 (DR1) galaxy clustering data, based on a joint full-shape analysis of the power spectrum multipoles and the bispectrum monopole using the ShapeFit framework. This is the follow-up of our previous work, in which we obtained for the first time constraints on the ShapeFit parameters using the bispectrum of DESI DR1. Here we present the first ShapeFit cosmological inference results using the bispectrum of DESI DR1. We recover values for the matter density parameter and Hubble constant of respectively $Ω_m=0.310\pm0.012$ and $H_0=[68.92\pm0.97]\,\mathrm{km\, s^{-1} Mpc^{-1}}$, consistent with previous results from the full DESI DR1 dataset that did not use the bispectrum signal. The inclusion of the bispectrum significantly tightens the constraints on the amplitude of fluctuations, reducing the error-bars in $\ln(A_s\times10^{10})$ by approximately 20\%, compared to using the power spectrum alone. We also explore extended cosmological models by performing fits for the evolving dark energy equation of state $w_0w_a$, and the sum of neutrino masses $\sum m_ν$. In these cases, we obtain constraints slightly larger than the ones from previous works from the DESI collaboration, due to not combining the full-shape results with other probes in all tracers. We find no strong evidence of deviations from standard $Λ$CDM, with the dark energy equation-of-state remaining within 2$σ$ from a cosmological constant $Λ$, and the neutrino mass being consistent with the normal hierarchy, $\sum m_ν<0.1\,[eV]$ at 95\% confidence limit. These constraints are broadly consistent with other DESI DR1 analyses, thus validating the robustness of the ShapeFit compression approach and the inclusion of the bispectrum for cosmological inference.

Figures (5)

• Figure 1: Baseline $\Lambda$CDM (68 and 95% confidence levels) joint constraints on the LRG+QSO samples from the DESI DR1 obtained with the ShapeFit compression method, complemented by the BAO postreconstruction data for the BGS, ELGs and Lyman-$\alpha$ tracers. In orange contours, the DESI DR1 official ShapeFit results KP5Desi, in purple the results obtained using the methodology of this work (see Section \ref{['sec:methodology']}) when only using the power spectrum, and in blue when the bispectrum is also added. The horizontal and vertical black dashed lines display the Planck 2018 $\Lambda$CDM central values. We note that including the bispectrum leads to constraints on $\ln(A_s\times10^{10})$ reduced by 20% with respect to the power spectrum-only approach, and that the power spectrum results from this analysis are completely consistent with the ones from the official DESI DR1 analysis.
• Figure 2: Constraints on $\Omega_m$ and $H_0$ with and without assuming an effective free sound horizon scale parameter $r_d^{\rm eff}$ obtained from the power spectrum and bispectrum ShapeFit results of DESI DR1. The blue contours correspond to the baseline ShapeFit $\Lambda$CDM results employing the standard pre-recombination physics $r_d(w_b,w_c,h,\ldots)$, which is a derived quantity from the model's cosmological parameters. The orange contours feature the same analysis settings, except that the sound horizon scale is effectively treated as a free independent parameter and marginalised over, $r_d^{\rm eff}$. Allowing the horizon scale to freely vary introduces a strong degeneracy between $H_0$ and $\Omega_m$, which follows $\Omega_m h^2$ degeneracy, and that significantly broadens the $H_0$ constraints. For completeness, we display in purple contours a similar analysis to the blue contours, but without using the shape parameters $m+n$ (i.e. the so-called standard compression or BAO+RSD).
• Figure 3: Joint 68 and 95% confidence levels on the dark energy equation-of-state parameters $w_0$ and $w_a$ from the DESI DR1 power spectrum and bispectrum ShapeFit analysis. The crosses indicate the obtained maximum a posteriori values from the two analyses, and dotted lines indicate the cosmological constant values ($w_0 = -1$, $w_a = 0$). The results are consistent with $\Lambda$CDM at the $2\sigma$ level, with no significant deviation in either parameter. The slight elongation of the baseline contour reflects the fact that most of the constraining power on the dark energy equation of state comes from the expansion history, which is more tightly constrained by post-reconstruction BAO analyses kp6bao. The constraints obtained with our methodology (in purple), which use post-reconstruction BAO data for the ELG, BGS and Lyman-$\alpha$ tracers and full-shape information for the LRG and QSO samples, are compared with the official DESI DR1 full post-reconstructed BAO+CMB results from kp6bao.
• Figure 4: Marginalized posterior for the sum of neutrino masses $\sum m_\nu$ from the DESI DR1 ShapeFit analysis (in purple) together with the constraints of DESI DR1 combined with the compressed Planck likelihood (in orange). This is consistent with previous results, both from the CMB collaboration2018planck and the DESI collaboration kp6baoKP7, which favour a sum below 100 meV.
• Figure 5: Results for different subsets of the ShapeFit compression. We show marginalized posterior distributions for the $\Lambda$CDM parameters $\Omega_m$, $H_0$, $\ln(A_s\times10^{10})$ inferred from the DESI DR1 data using the joint power spectrum-bispectrum analysis. The blue contours correspond to the full ShapeFit analysis (which are therefore identical to the blue constraints in Figure \ref{['fig:PvsB']}). The orange contours use only information from the combined shape parameter $m+n$, while the purple constraints are obtained from the standard compression approach (using the parameters $\alpha_\parallel,\alpha_\bot,f,\sigma_\textrm{s8}$).